Dear colleagues,

I already have some thoughts on this but I ask to get some feedback about what you think.

Together with colleagues, I am conducting a meta-analysis that aggregates simple bivariate correlation coefficients. The independent variable is a firm characteristic, the dependent variable a performance measure. While I am the "analyst", others collect the data.

Yesterday I saw that we have a substantial amount of panel studies in our data, where the correlation matrix (that is the primary input) was estimated across the whole panel and not pooled on the firm level. I would suggest that these cannot be compared to pooled correlations, that we require, correct?

Thanks for some thoughts,

Holger